lable at ScienceDirect

Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531

Contents lists avai

Journal of Rock Mechanics andGeotechnical Engineering

journal homepage: www.rockgeotech.org

Full length article

A borehole stability study by newly designed laboratory testson thick-walled hollow cylinders

S.S. Hashemi*, N. Melkoumian, A. TaheriDeep Exploration Technologies Cooperative Research Centre, School of Civil, Environmental and Mining Engineering, University of Adelaide, Adelaide,Australia

a r t i c l e i n f o

Article history:Received 25 March 2015Received in revised form4 June 2015Accepted 9 June 2015Available online 30 July 2015

Keywords:Real-time monitoringExperimental investigationThick-walled hollow cylinder (TWHC)Poorly cemented sand formations

* Corresponding author. Tel.: þ61 426111280; fax: þE-mail address: shashemi@civeng.adelaide.edu.auPeer review under responsibility of Institute o

Chinese Academy of Sciences.1674-7755 � 2015 Institute of Rock and Soil MecSciences. Production and hosting by Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.jrmge.2015.06.005

a b s t r a c t

At several mineral exploration drilling sites in Australia, weakly consolidated formations mainly consistof sand particles that are poorly bonded by cementing agents such as clay, iron oxide cement or calcite.These formations are being encountered when drilling boreholes to the depth of up to 200 m. To studythe behaviour of these materials, thick-walled hollow cylinder (TWHC) and solid cylindrical syntheticspecimens were designed and prepared by adding Portland cement and water to sand grains. The effectsof different parameters such as water and cement contents, grain size distribution and mixture curingtime on the characteristics of the samples were studied to identify the mixture closely resembling theformation at the drilling site. The Hoek triaxial cell was modified to allow the visual monitoring of graindebonding and borehole breakout processes during the laboratory tests. The results showed the sig-nificance of real-time visual monitoring in determining the initiation of the borehole breakout. The size-scale effect study on TWHC specimens revealed that with the increasing borehole size, the ductility ofthe specimen decreases, however, the axial and lateral stiffnesses of the TWHC specimen remain un-changed. Under different confining pressures the lateral strain at the initiation point of boreholebreakout is considerably lower in a larger size borehole (20 mm) compared to that in a smaller one(10 mm). Also, it was observed that the level of peak strength increment in TWHC specimens decreaseswith the increasing confining pressure.� 2015 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by

Elsevier B.V. All rights reserved.

1. Introduction

Borehole stability analysis is an important issue for researchersin the field of geotechnical, mining and petroleum engineering.Several borehole instability problems during or after the comple-tion of drilling have been reported by a number of explorationcompanies in Australia. Many of these problems were reported indrilling projects in poorly cemented sand formations at the depth ofup to 200 m beneath the ground. The sand production problem, asit is known, has also been observed in weakly bonded sandstoneswhere the debonding of sand grains can be triggered by fluidpressure and induced stresses leading to the failure of sandstone atthe borehole wall (Geertsma, 1985; Perkins and Weingarten, 1988).The strength of a granular material formation is generated mainly

61 883134359.(S.S. Hashemi).f Rock and Soil Mechanics,

hanics, Chinese Academy ofrights reserved.

by a natural cementing agent that bonds sand grains together (Al-Awad et al., 1999).

In recent decades, a number of numerical and experimentalstudies have been conducted on borehole stability in different rockformations. Gough and Bell (1982) showed that, in a large numberof vertical oil wells, the orientation of consistent breakouts co-incides with the direction of the regional minimum horizontalprincipal stress. Numerical studies on poorly cemented sand for-mation by Hashemi et al. (2014a, b) revealed that breakage of weakbonding between sand particles causes instability and sand grainsremain intact in the case of a borehole failure. In laboratory con-ditions it was shown that borehole breakouts grow mainly throughradial penetration into the rock mass without any circumferentialextension (Lee and Haimson, 1993). Haimson and Song (1993)conducted laboratory tests on two varieties of Berea sandstoneswith 17% and 22% porosity, respectively, and revealed two distinctbreakout patterns directly related to the microstructures of theserocks and their different modes of failure. Microscopic observationby Hagin and Zoback (2004) revealed that the fundamentalmechanism behind sand production is the growth of small,opening-mode and splitting cracks oriented parallel to thetangential stress, starting very close to the borehole wall andpropagating deeper into the matrix with increasing stress. Similar

Silt and sand

Clay

Poorly cemented sands

Hard rock

Fig. 1. Vertical geological cross-section near Burra, South Australia (Hashemi et al.,2014a, b).

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531520

borehole instabilities have also been documented by many otherauthors (Fischer et al., 1978; Ewy and Cook, 1990a, b; Chang et al.,1997; Al-Ajmi and Zimmerman, 2006).

Unconfined compressive strength triaxial and thick-walledhollow cylinder (TWHC) tests are two of the most useful labora-tory experiments (King, 1912; Robertson, 1955; Hoskins, 1969;Adeyanju and Olafuyi, 2011). The geometry of the TWHC allowsthe application of various load path combinations to simulate stressconditions around boreholes. The classical hollow cylinderapproach is not well suited to investigating the stability of poorlycemented formations because, during or subsequent to these tests,the instability of the weak sandstone arch and the grain debondingprocess in sandy formations cannot be captured. The hollow cyl-inder tests on an intact rock sample focus mainly on rupture phe-nomena such as shear and tensile failure (Hall and Harrisberger,1970; Tippieand and Kohlhaas, 1973; Cleary et al., 1979).

In this study, a series of newly designed laboratory testsinvolving real-time monitoring of the development of breakout inan unsupported borehole was conducted. The paper aims to pro-vide a more accurate representation of the actual behaviour ofpoorly cemented sands, which will be invaluable in designingappropriate borehole support systems. The tests were conductedon specimens of poorly cemented sands prepared in laboratory andthe effects of different mixture characteristics (i.e. proportion ofsand, cement and water) on their mechanical behaviours werestudied by conducting compression tests on solid and hollow cy-lindrical specimens.

2. Drilling field investigation

Exploration boreholes are usually drilled to uncover potentialfuture mine sites. In many cases, drilling is undertaken throughpoorly cemented sand formations. Generally, the boreholes are250e300 mm in diameter and 50e200 m in length, depending onthe underground conditions in Australia. This study focuses onsolid and TWHC laboratory test specimens based on disturbedsamples collected from a problematic drilling site at Burra, SouthAustralia. At this site the sediment above bedrock is heterogeneous,with the shallower layers composed of silt and fine sand and thedeeper layers transition to dark grey plastic clay. The problematic,poorly cemented sandstone underlies this clayey layer and consistsof sand particles with a weak cementation due to the presence ofiron dioxide, clay and calcite (Fig. 1). Quartz grains are mostly fineand sub-angular with random orientations. The yellowish-greyspecimens were prepared in laboratory so that their fabric closelyresembles that of the poorly cemented sands at the drilling site.

To drill a borehole at this site, different drilling methods havebeen tried to minimise the risk of borehole failure. In some cases,drilling mud has been used to maintain an open borehole duringdrilling. However, to conduct further investigation, some boreholesneed to remain open for several months after drilling. The air coredrilling and reverse circulation drilling methods were used in thesecases. These are dry drilling methods and have been conventionallyapplied to drilling through soft ground in Australia. The cuttings areconveyed to the surface and pass through the sample collectionsystem from where they are collected. Drilling continues with theaddition of rods to the top of the drill string. When the drillingstring reaches the poorly cemented sand layer, the borehole maycollapse if the bonding between sand particles is not strong enoughto provide stability. According to the reports from drilling company,the main factors affecting borehole instability include the lowstrength of poorly cemented sands which cannot sustain theexisting in-situ stress after drilling, and, in few cases, fluid flow dueto a confined aquifer near the borehole collapse zone.

3. Thick-walled hollow cylinders (TWHCs)

Hollow cylinder specimens were first used in the early 20thcentury when the importance of adopting a realistic model wasidentified for an underground opening. The opening was located ata depth of 9.5 km and susceptible to collapse due to high in-situstresses (Hoskins, 1969). Since then, a wide range of experimentalinvestigations involving hollow cylindrical specimens has beenconducted. Robertson (1955) studied the effect of the inner-to-outer diameter ratio on the strength of various rocks. King (1912)analysed the system of fractures that might develop duringcompressive tests on hollow cylinder specimens under differentstress states. Bridgman (1952) performed hollow cylinder testsunder different loadings. Pomeroy and Hobbs (1962) examined thestrength of hollow cylinder coal specimens. Mazanti and Sowers(1966) studied the behaviour of hollow cylinder granite speci-mens and the effect of the intermediate principal stress (s2) ontheir strength. Ewy et al. (1988) studied the deformation andfracture development in a hard rock around a borehole using TWHCtests. These and other work involving TWHC tests show that theTWHC configuration tests are well suited for identifying andinvestigating both the macro- and micro-properties of differentrock types.

Depending on the thickness of a hollow cylinder specimen,stresses that develop in its wall due to the application of uniformstresses can be analysed by two different methods. In a TWHCspecimen, the wall thickness, t, which is much smaller in compar-ison to its inner diameter, Di (i.e. Di > 20t), the stresses thatdistribute across the specimen walls can be considered almosthomogeneous and uniform. This assumption is not fully acceptablein the case of a thick-walled hollow specimen, the wall thickness ofwhich is larger in comparison to its inner diameter. In the lattercase, stress distribution is not homogenous in the specimenwall. Inmany textbooks (e.g. Obert and Duvall, 1967; Jaeger et al., 2007)closed form solutions based on the linear theory of elasticity forcalculating stresses and strains in both thin- and thick-walledhollow cylinder specimens were presented. For a TWHC with aninner diameter of Di, outer diameter of Do and length of L, subjectedto axial force (F), uniform internal stress si and external stress so, theprincipal stresses at any point at a radial distance r from the centreof the specimen can be presented in cylindrical coordinates asfollows (Jaeger et al., 2007):

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 521

sqq ¼ soD2o � siD2

i

D2o � D2

i

þ ðso � siÞD2i D

2o

4r2�D2o � D2

i

� (1)

srr ¼ soD2o � siD2

i

D2o � D2

i

� ðso � siÞD2i D

2o

4r2�D2o � D2

i

� (2)

sz ¼ 4F

p�D2o � D2

i

�þ siD2i

D2o � D2

i

(3)

where sqq; srr and sz are the tangential, radial and axial principalstresses, respectively, which are assumed to be uniformly distrib-uted over the specimen on top and bottom end surfaces (Fig. 2).

4. Experimental study

Laboratory test facilities that were used in the current experi-mental testing comprised the following components:

(1) Specimens that were both of a reasonable diameter and of aborehole wall thickness satisfying the TWHC theory condition(i.e. Di < 20t) were used. Hence, a HQ Hoek triaxial cell of63.5 mm in diameter and 127 mm in height was utilised.

(2) A servo-controlled axial loading system of 100 kN loading ca-pacity with 0.1 N accuracy was used for applying vertical stressto the specimen.

(3) Although the Hoek cell was originally designed to apply highconfining pressures to hard rock specimens, the hydraulicpressure gauge was modified to allow measuring the confiningpressure at very low amounts. An automatic hydraulic machinewas used in conjunction with a relief valve and a pressuregauge for applying and maintaining the external confiningpressure. Since the maximum confining pressure adopted inthe current tests was low (6 MPa), a pressure gauge with anaccuracy of 0.01 MPa was used.

The TWHC specimens, consisting of poorly cemented sands,cannot be retrieved from the cell after the destructive tests due tothe development of a large number of macro- andmicro-cracks andthe debonding of sand particles. In addition, the specimen isattached to the membrane and completely crumbles when moved.Thus, it is not possible to retrieve poorly cemented sand specimens

Fig. 2. In-situ stresses on an element at a radial distance r from a drilled boreholecentre in polar coordinates.

and investigate the failure mechanism of borehole after the test. Toaddress this issue, the triaxial cell was modified to allow simulta-neous capturing of the borehole failure mechanism and the processof sand grain debonding at each time step and at different stresspaths. A micro-camera with a 225 pixel per inch (ppi) resolutionwas installed inside the hollow platen to record the process of sanddebonding and borehole breakout. The micro-camera was con-nected to a personal computer to record the borehole conditionsthroughout the test. A 60-channel data acquisition system wasconnected to two additional personal computers for recording andstoring data.

Initially, three PVC moulds with a slot on the circumferencewere manufactured to prepare the specimens. However, during thecompaction process, the body of the mould bulged and thisdeformation prevented the specimens from obtaining a uniformcylindrical shape. An additional problem was observed during de-moulding. Since the specimens were very weak, deformation ofthe mould to release the specimen imparted damage to it. Toaddress this issue, steel moulds were designed and manufactured(Fig. 3a). The dimensions of the moulds were 127mm in length and62.5 mm in diameter. Two removable steel dowels with the di-ameters of 25 mm and 10mmwere used together with the moulds.These dowels were embedded in the mould to create the boreholein the specimens. As shown in Fig. 3a, to avoid damaging thespecimen during de-moulding, the moulds comprised two half-cylinders that were joined together by two screws. In addition,the dowels were wrapped in a plastic film (Mayla plastic) and theinner surface of moulds was lubricated with grease (containing apetroleum jelly mixture and stearic acid) which would not pene-trate into the mixture. The bottom platens of the moulds werecoated with an anti-rust paint to diminish the effect of lubrication.It was observed that unlike the concrete, the strength of the poorlycemented sand specimens was strongly influenced by the mouldconditions.

Kongsukprasert (2003) showed that the strength of poorlycemented sand samples is a function of the density of the mixture.Since there was insufficient space between the mould’s inner walland the internal dowel to facilitate compaction, a tool was designedand manufactured to uniformly compact the mixture, as shown inFig. 3b. Each specimen was compacted in three separate layers ofequal thickness (42 mm). The compaction energy for each impactwas maintained constant and equal to 0.35 N m/cm3, which wascalculated based on the applied force of the manufacturedcompactor. Before placing the next layer, the surface of the previ-ously compacted layer was scarified to increase the interlockingbetween successive layers. To minimise the bedding error effect forthe very top layer, a collar was used allowing this layer to achieveconditions similar to those of the lower layers (see Fig. 3a). To avoidthe initial setting of the cement for all specimens, the compactiontime was strictly maintained at 20e30 min. The compaction beganwhen water was first added to the mixture and was completedwhen the final layer was compacted.

Two sets of cylindrical platens were manufactured from hard-ened steel and were hardened prior to crushing and lapping. Theplatens were designed using the finite element analysis softwareABAQUS 6.11 and loading steps similar to those applied during testconditions were applied to the model (Fig. 4). According to theresults of simulated platens, the strain of the platens was less than0.01% with the application of the maximum load of 100 kN. This isfar greater than the predicted strength of the specimens. Eachplaten was designed with a tapered hole, 35 mm at the top and25 mm or 10 mm at the bottom based on the TWHC specimen holesize. In addition, a small base was fitted to the platen to fix thecamera in the position at a small distance above the specimen. Thebottom platens were fitted with a uniform cylindrical hole of

(a) (b)

D62.00

D100.00

Hollow bar

Plan view

Top plate

Lower plate

I.D26.00

Intternal dowel Colllar

Fig. 3. Apparatus developed to prepare the specimens: (a) Special moulds for preparing TWHC specimens; and (b) Special device for compacting the mixture in the moulduniformly (sizes in mm).

Section a-aPlan view

a a

8

3R12.5R31.76

30

25

25

160

63.6

Micro-camera

Micro-LED

30 27

Fig. 4. The designed and manufactured top and bottom platens (unit: mm).

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531522

20 mm or 10 mm in diameter, which can be fitted to the bottomram of the loading machine. These platens were also simulated inABAQUS. Sand particles, which were debonded from the boreholewalls during the test, were allowed to fall onto the bottom platen.

Since the specimens were weak enough to simulate the poorlycemented sand at the drilling site, they were vulnerable to distur-bance prior to loading. To avoid applying the weight of the triaxialcell to the specimen during the test, a wooden base was manu-factured to support the weight of the triaxial cell. Thus, there wasno need to apply an external pressure on the specimens to hold thetriaxial set before transferring it to the loading machine.

4.1. Test procedure and setup

The cell was placed on the wooden base and the bottom hollowplaten mounted by the loading machine manually ensured that itcan move freely inside the Hoek cell. A thin black membrane wasused at the end of the bottom hollow platen to ensure that the LEDlights of the camera are not reflected during video recording. Bothends of the specimen were then levelled by applying a thin layer ofdental paste and, after it had been set, lubricated with a specialgrease to reduce the friction between the platen and the specimen,thus limiting stress concentration and bedding error. During thesetting of the dental paste, the specimen was wrapped in a plasticfilm to avoid excessive drying of its surface, which could result inthe loosening of particles. After the dental paste had been set,

precise measurements of the specimen’s weight, height anddiameter were taken at three different points bymeans of a calliper.

Measuring the dental paste deformations at the top and bottomsurfaces of the specimen during the test revealed that the dentalpaste could withstand, without any noticeable deformation, a forceof up to 100 kN, which is far greater than the strength of the poorlycemented specimen. The results from the preliminary tests on thespecimens without capping were not reproducible due to the sig-nificant inconsistency in the results.

The top hollow platen was positioned on the specimen and twospherical platens were placed on top of it to ensure that the verticalstress was uniformly applied to the specimen. Pairs of axial andlateral strain gauges were used to measure local deformations onthe specimen. Two linear-variable differential transformers (LVDTs)were installed between the top and bottom rams of the loadingmachine to measure axial displacement externally. Prior to the test,the uppermachine ramwas brought to the edge of the top platen toset the offsets and 5 N was applied to ensure the contact betweenthe top ram and the hollow platen. The captured image of themicro-camera was checked to ensure that the focal length of thelens was on themiddle of the specimen hole and the position of theLEDs was controlled to ensure the borehole illumination was suit-able for recording. In the first stage of loading, the vertical andconfining stresses were increased simultaneously up to a certainstress level, which simulates the hydrostatic condition on thespecimen boundary. Then, in the second stage, the sample wassubjected to vertical compression at a constant displacement rate

Fig. 5. Proctor compaction test results of silica sand with and without Portland cementpowder (wc ¼ 6%). gd is the dry density (g/cm3), and ww is the water content (%).

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 523

of 0.07 mm/min. The effects of various strain rates (0.02e0.1 mm/min) were also examined. No significant change was observed inthe strength and strain behaviours of the specimens within thisstrain range. Data were recorded at 0.5 s time intervals.

As has been reported by various drilling companies, time is a keyfactor in predicting the borehole stability after drilling or whenwithdrawing the drilling rods from the borehole to change thedrilling bit. The video capture software was synchronised with thedata acquisition system to facilitate the observation of the sandparticle debonding process in parallel with the stress and strainmeasurements.

Several sets of synthetic mixtures of poorly cemented sandswere prepared and tested with different cement-to-sand ratios (byweight), wc, coarse-to-fine sand particle ratios, d, and grain sizeranges.

4.2. Grain size distribution

To determine typical grain size distributions, samples werecollected from the depth of up to 100 m at the drilling site in Burra,South Australia, and sieve analyses were performed using ASTM C-136 calibrated sieves plus pan. The particle size distribution wasfound to be almost uniform fine grained. Based on these sieveanalysis results, natural silica sands (99.6% of silica) of two differentgrain size ranges, closely resembling the ones at the drilling site,were selected for preparing the synthetic mixtures in the labora-tory. For these sand grains, the mean diameter (D50) for particlesizes of 0.425e1.4 mm (termed as “coarse”) was 0.56 mm and, forgrain sizes of 0.125e0.355 mm (termed as “fine”), was 0.20 mm.The coefficient of uniformity (Cu ¼ D60/D10) for the coarse and finesand grains was 1.452 and 2.268, respectively. The sands weresieved and packed into plastic bags, about 3 kg in each, to preservethe natural moisture content of the sands. The density of the fineand coarse sand particles was 1.47 g/cm3 and 1.59 g/cm3,respectively.

4.3. Water content

According to Gueguen and Palciauskas (1992), water contentshould be kept at a minimum value to avoid segregation of thecement and sand grains. Different water contents were consideredto create the most suitable mixture. To determine the optimumwater content for the mixture, standard Proctor compaction testswere conducted (Fig. 5). These tests were performed on mixturesboth with and without cement. Fig. 5 shows that the optimumwater ratio evaluated by using compactor hammer with energy of0.55 Nm/cm3 was achieved at 9.7%e10.3%. Therewas no significantchange in optimum water content and density in the case of thesand and cement mixture or of the sand grains only.

4.4. Cement content

Portland cement type II (specific gravity Gs ¼ 3.15) was used forpreparing the TWHC specimens based on previous studies on themechanical properties of weakly cemented sands (Saidi et al.,2003). The cement powder used in the current study was from asingle bag and it was kept in a sealed and airtight containerthroughout the laboratory studies.

For preparing poorly cemented sands in laboratory conditions, awide range of wc values have been suggested by different re-searchers (Alsayed, 1996; Kongsukprasert, 2003; Saidi et al., 2003,2005). Kongsukprasert (2003) used a maximum of 2.5% of Port-land cement, whereas Saidi et al. (2003) used 9%e18%. In addition,Gueguen and Palciauskas (1992) stated that the minimum wc isreached when the coarse-to-fine sand particle ratio, d, is equal to

1.5. Saidi et al. (2005) showed that a very small amount of Portlandcement will increase the strength and stiffness of cemented sandmixtures if deposited at grain-to-grain contacts. However, based onthe authors’ observations, using sands of the above-mentionedgrain size distribution, for wc � 2.5%, the specimens could not besuccessfully de-moulded after the curing time (2e12 days).

A number of mixtures with different values of wc and d wereprepared and examined to achieve the apparent mechanicalbehaviour of the drilling site samples. For the mentioned particlesize distribution range, values of wc ¼ 6%, 7% and 8% were selected.According to the test results, since the grain size distributionsignificantly affects the strength of the specimens, it is not possibleto create poorly cemented sand samples by using a fixed wc valuefor different particle size distributions.

4.5. Coarse-to-fine sand particle ratio (d)

The cement content, wc, and the coarse-to-fine sand particleratio, d, were varied to obtain the desirable mixture. In this study,since the cement powder particle is finer than the sand grains, thecured Portland cement is assumed to be a continuous phase despitethe fact that it includes micro-granule and micro-porosity (Saidiet al., 2005). The specimen contains three separate phases: sand,cement and macro-porosity larger than the sand grain size (Ashbyand Jones, 2014). Scanning electron microscope (SEM) photographsconfirmed that macro-porosities were often larger in size than thesand particles themselves (Fig. 6a). According to Saidi et al. (2005),the minimum macro-porosity is achieved when d is around 1e3.Also, based on the collected samples from the drilling site, thecured mixture should be weak enough to allow sand particles todebond when the surface is scratched with a finger nail. Therefore,different values of d were examined for a constant wc to determinethe most suitable d. Finally, d ¼ 1 was chosen based on the testoutcomes which are discussed later.

4.6. Curing time

Various curing times have been suggested in previous studies(e.g. Kongsukprasert, 2003; Saidi et al., 2003). In the currentresearch, curing times of 2e12 days were examined. The curingtime includes the curing of the mixture, both in themould and afterde-moulding. In the case of less than 5 days curing, the specimens

Table 1Properties of the prepared poorly cemented sand specimens.

wc

(%)Porosity,n (%)

Tangent elasticmodulus,Etan (GPa)

UCS(MPa)

Poisson’sratio, n

Coulombparameters

Bulkdensity,r (kg m�3)

c (MPa) f (�)

6 26 � 3 2.262 0.62 0.294 1.11 29.67 19547 26 � 2 2.465 1.05 0.250 1.38 30.06 19698 26 � 2 3.360 1.90 0.247 1.59 29.67 1974

Fig. 6. (a) SEM photograph showing that macro-porosities are larger than sand par-ticles in the cemented sand mixture. (b) Borehole breakout occurring in the area whichseems to still contain some moisture.

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531524

could not be successfully de-moulded. Preliminary tests wereperformed on TWHC specimens with total 5 days curing time.Careful examination of the video taken during testing and of thecross-sections of the tested specimens (Fig. 6b) revealed that allboreholes had failed in a zone that seemed to have retained somemoisture and had not yet completely dried. After further studies,based on the uniaxial compressive strength (UCS) test results, 5-day curing was deemed to be the optimal time for curing inside themould when the specimen was compacted at atmospheric pres-sure. After removal of the mould, each specimen was wrapped byplastic film, placed in an airtight plastic container and left to furthercuring for another 3 days under atmospheric pressure and at aconstant water content. To ensure that 26%e29% porosity wasachieved, no external pressure was applied to the specimens whilstcuring.

5. Results and discussion

In order to determine the properties of the mixture, variouspreliminary tests were conducted on solid and TWHC specimens.The results of these tests can be used to compare other drillingfields and ground conditions to those examined in the current

study. Preparation of the mixture began with wc ¼ 1% using onlyfine grained sands. Table 1 presents the details of the physico-mechanical properties that were observed from the uniaxial andtriaxial tests. To determine the mechanical properties of the spec-imens, a number of solid cylindrical specimens with different sandgrain size distributions, Portland cement and water contents wereprepared.

5.1. Effect of coating on the strength of TWHC specimens

Based on the preliminary test results, foil strain gauges andexternal LVDTs were selected to measure the axial and lateralstrains during triaxial testing. To identify the most suitablecoating for poorly cemented sands, three specimens with thesame characteristics were prepared. Four strain gauges wereplaced on each specimen with different underlining conditions(i.e. polymer, plastic bond, no lining). The results from the straingauges using the plastic bond lining and the specimen withoutlining showed irregularities in the stressestrain diagrams, but thestrain gauges placed on the polymer coating showed good andreproducible results and the cohesion of the polymer and spec-imen continued to be preserved after the test. Thus, polymercoating was deemed to be the best suited option for the subse-quent triaxial testing.

In addition, the strengthening effect of the polymer on thepoorly cemented sand TWHC specimens was considered. Videorecordings that were taken inside the borehole of the preliminaryTWHC tests showed that the debonding of sand particles in thespecimens initiated in the zone out of the applied polymer location(e.g. top and bottom sections of the borehole, see Fig. 7a). In thetests without coating, debonding often began uniformly in themiddle of the specimens. However, using polymer coating isessential for obtaining accurate readings from the strain gauges, asdiscussed earlier. To address this problem, in subsequent tests athin layer of polymer was applied along the length of all thespecimens symmetrically as is shown in Fig. 7b. The results showedthe significance of using the camera in identifying the factorsaffecting the process of sand debonding from the borehole wall.

5.2. UCS tests on solid specimens

5.2.1. Effect of particle size distribution in UCS testsAs mentioned above, two sets of sands, namely fine (0.125e

0.355 mm) and coarse (0.425e1.4 mm) grained sands, wereexamined in this study. Three sets of UCS tests were performed onsolid specimens with the same values ofwc ¼ 8% andww ¼ 10%, butwith different coarse-to-fine sand particle ratios (d) (i.e. 0.14, 0.33and 1). The specimens were left to curing for 5 days at atmosphericpressure with the same curing method that was discussed earlier inSection 4.6.

Fig. 8a shows that the influence of d on the peak strength andpre-peak stiffness is considerable. It is evident that, whilst main-taining wc constant (i.e. at 8%), the strength and stiffness of thespecimens increase as the proportion of coarse sand particles (i.e. d)

Fig. 7. (a) Borehole breakout occurring in an area out of strain gauge coating; and (b) polymer application pattern to minimise the effect of coating on the strength of the specimens.

Fig. 8. Results of UCS tests. (a) Effect of different d values on the strength of poorlycemented sand specimens; (b) Effect of different wc values on the stressestrain dia-grams and (c) on the volumetric strain, where εa is the axial strain.

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 525

rises and the specimens become more brittle in the post-peakregime. The strength of the mixture increases significantly asd grows because the coarse sand particles exhibit greater resistanceto rotation in the specimen matrix than the fine particles. There-fore, when the number of the coarse grains increases, the rotationof particles becomes more difficult due to the improved level ofinterlocking between the particles after compaction and the globalcompressive strength of the mixture is increased. In the case ofhigher d, the other reason behind the mixture strength growth isthe decrease in the special surface of sand particles in the wholematrix which enhances the influence of cementation in the spec-imen. Therefore, by increasing the sand grain size, it is possible toobtain a higher strength mixture for a certain wc value. This con-firms the findings by other researchers (Kongsukprasert, 2003;Saidi et al., 2003, 2005) who used different ranges of wc to createcemented sand specimens. Also, the level of compaction improvedwith the increase of d up to a certain level and thus the cementingagent could bond more sand particles in a unit volume.

It should be noted that increasing d by more than 1 was notconsidered in this research, because the aim was to test particlesizes similar to those observed at the Burra drilling site.

5.2.2. Effect of cement content in UCS testsTo study the effect of cement content (wc), three different values

of wc were considered, namely 6%, 7% and 8% under the sametesting conditions. To examine reproducibility of the results, spec-imens with 7% and 8% of cement content were tested twice. Fig. 8bpresents the stressestrain diagrams of the specimens with a drydensity of 1.81 g/cm3 after 7 days curing.

From the UCS test results, the following trends can be observed:

(1) The maximum strength, smax, and pre-maximum stiffness in-crease aswc rises. This outcome is in agreement with Saidi et al.(2003) and Kongsukprasert et al. (2005) who used other rangesfor wc and d.

(2) The post-peak stressestrain trend of the specimens with lowervalues of wc exhibits greater ductile behaviour. As shown inFig. 8a and b, the UCS is affected to a greater extent bywc than d.

The effect ofwc on the volumetric strain is presented in Fig. 8c. Itcan be observed that as wc increases the specimens experiencegreater axial and volumetric strains prior to failure. In addition, forlower wc values, dilation is evident in the specimens after a limitedamount of compression. However, with increasing cementation thelateral strain decreases and the volumetric strain is affected mostlyby the axial compression strain. The axial strain increased by about83% and 84% at εv ¼ 0 withwc increasing from 6% to 7% and from 7%to 8%, respectively.

Furthermore, the effect of water required for cement hydrationwas considered by increasing the water content in small

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531526

increments within the range of 5%e12%, covering the dry, opti-mum, and wet sections of the optimum water diagram.Kongsukprasert et al. (2005) showed that, for the cemented sandmixture, the added water cannot be entirely used for cement hy-dration. The total water content of the mixture comprises of a partthat serves to hydrate the cement and another part is absorbed bythe grains at the time when water is added to the mixture. Inaddition, Chen and Wu (2013) suggested that the degree of hy-dration increases as the curing time and water-to-cement ratio ofthe mixture rise. However, they showed that excessive water con-tent increases the total porosity and results in strength reduction.When the sand grains are not in the saturated-surface dry (SSD)mode, part of the water will be absorbed by the grains, while theremaining part covers the particles’ surfaces. Specimens with lessthan 6% water content could not be de-moulded after even 10 and12 days of curing time, which inferred that, due to the high specificsurface of the sand particles, a minimum of 6% water content wasrequired to achieve hydration. The results showed that, for a spe-cific density of a compacted mixture, the highest strength is ob-tained with the water content at or around the optimum one, andthe strength is significantly reduced when the water content is lessthan the optimum value. As mentioned earlier, the optimumwatercontent was determined by standard compaction tests and wasfound to be almost 10%.

5.3. Triaxial test results on solid specimens

Triaxial tests on solid specimens were performed to determinethe shear failure properties of the specimens under different stressconditions. Table 2 shows a summary of the final values of d,wc andww which were identified by preliminary tests, where sconf is theconfining pressure.

Fig. 9aec presents the results of the triaxial tests on the solidspecimens with different wc values. As shown in Fig. 9, an increasein confining pressure results in an increase in peak strength.However, the confining pressure has a minimal effect on thestiffness of the specimens. Also, as it can be seen from Fig. 9, thismaterial exhibits strain hardening behaviour, i.e. continuous in-crease in the deviatoric stress with axial strain. Thus, the stresscorresponding to 1% of axial strain was considered as themaximum strength of the specimens. The results show that thematerial behaves nonlinearly even in the early stages of theloading process as was seen by the UCS tests earlier. Paterson(1967) showed that ductility increases with increasing s3. Fig. 9shows that the ductility of the specimens at lower confiningpressures is less significant and increasing the confining pressureresults in the transition from brittle to ductile behaviour. Ac-cording to the results, for wc ¼ 7% and 8%, the elastic strain re-mains relatively constant due to an increase in confining pressure.Higher confining pressures (e.g. >6e6.5 MPa) were examined forspecimens in each of the three categories and the results showedthat strain gauges on the specimens failed at these higherconfining pressures. This problem often occurs due to the sandysurface and weak matrix of this material.

Table 2Schedule of the triaxial tests conducted on solid and poorly cemented TWHC sandspecimens.

Type ofspecimens

wc (%) d Boreholesize (mm)

sconf(MPa)

Displacementrate (mm/min)

ww (%) Curingtime (d)

Solidcylinder

6, 7, 8 1 e 1e5 0.07 10 8

TWHC 6, 7, 8 1 10, 20 1e5 0.07 10 8

Fig. 9. Deviatoric stress versus axial and lateral strain behaviours of solid specimenssubjected to triaxial testing: (a) wc ¼ 6%; (b) wc ¼ 7%; and (c) wc ¼ 8%.

5.3.1. MohreCoulomb strength parametersMohreCoulomb criterion is one of the most commonly used

failure criteria in borehole stability problems (Jaeger et al., 2007).Coulomb suggested that shear failure will take place when theshear stress is equal to the sum of cohesive shear strength and theproduct of the coefficient of internal friction and normal stress

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 527

across the fracture plane. Later, Mohr asserted that the relationshipbetween the normal and shear stresses along the failure plane isnonlinear:

s ¼ f ðsnÞ (4)

where f is a function that can be derived empirically.To explain the behaviour of the material by the Coulomb failure

criterion, the cohesion, c, and the angle of internal friction, f, weredetermined. As expected, increasing the cement content produceshigher cohesion between the sand grains, and based on the MohreCoulomb criterion, c (cohesion) changes for different cementvalues. The c and f values are presented in Table 1. For different wcvalues, the derived f values show that the angle of internal frictiondoes not change dramatically due to the increase in the cementcontent. This is mainly due to the grain size distribution not beingchanged in the specimens tested.

Fig. 10. Volumetric strain versus axial strain results from triaxial tests on solid andTWHC (10 mm borehole) specimens: (a) wc ¼ 6%; (b) wc ¼ 7%; and (c) wc ¼ 8%.

5.3.2. Volumetric strain resultsLateral strain was measured during the triaxial tests on solid

specimens to study the effect of confining pressure and cementcontent on the volumetric strain of the prepared specimens. Fig. 10shows the average values of the volumetric strains calculated basedon average values of lateral and axial strain gauges for solid andTWHC specimens. As can be seen in this figure, at lower confiningpressures (i.e. 1 MPa and 2 MPa) the solid specimens began tocontract from the top and bottom sections first and after 0.55% axialcompression strain, it diverted to lateral dilation. Increasing theconfining pressure kept the specimen in a more contraction modeand dilation began at higher axial strain values. With an increase inthe confining pressure, the lateral strength of the specimens willrise and the lateral dilation will be lower for the same axial strain.As mentioned earlier, after the elastic phase in the stressestraindiagram, the solid specimens exhibited strain hardening behaviour.Thus, for confining pressures greater than 2 MPa, the solid speci-mens remained in contraction mode and never conjugated thehorizontal axis (i.e. the zero volumetric strain) until the maximumstrength which was assumed equivalent to 1% of axial strain. Theresults also show that the effect of varying the cement contents onthe volumetric strain is less significant than altering the confiningpressure. For instance, at confining pressures of 1 MPa and 2 MPa,increasing the wc value from 6% to 8% does not result in anydivergence in the behavioural trend from contraction to dilation inthe specimens.

Fig. 9 also presents the deviatoric stress versus lateral strainbehaviour for solid specimens. It shows that applying higherconfining pressure slightly increases the stiffness of the material inthe lateral direction. In addition, the specimens exhibited morebrittle behaviour at low confining pressures and, with increasingconfining pressure, thematerial response becamemore ductile. It isworth mentioning that the specimens underwent lower lateraldeformation by increasing the wc value at a constant ww, d andconfining pressure.

5.4. Triaxial test results on TWHC specimens

Triaxial tests on TWHC specimens were conducted to investi-gate the borehole breakout and failure properties of the syntheticspecimens under different stress conditions. Real-time videorecording helped to determine the initiation and direction ofborehole breakout and to locate the sand debonding on the bore-hole wall. The main TWHC tests were conducted for three differentcement contents (wc) and for two different borehole diameters:10 mm and 20 mm.

Fig. 11 presents the results of the triaxial tests on 10 mmdiameter borehole TWHC specimens for three different wc values(6%, 7% and 8%) and various confining pressures. As shown inFig. 11a, increasing the confining pressure from 1 MPa to 3 MPasignificantly enhances the level of peak strength and strain energyin the specimens. It should be mentioned that the pre-peak stiff-ness does not change dramatically with an increase in the confiningpressure for a certainwc. However, for sconf> 3MPa, the level of thestrength increment is lower than that for the previous states. Forconfining pressures higher than 4.5 MPa, the first stage of the test(i.e. applying hydrostatic stress to the boundary of the specimens)could not be completed and the borehole collapsed immediately.

Fig. 12 shows the process of the borehole failure prior toapplying the deviatoric stress. However, the axial strain gaugescontinued to function even after failure. This emphasises the sig-nificance of using the micro-camera during the borehole stabilitystudies.

Fig. 11. Deviatoric stress versus axial and lateral strain behaviours of 10 mm boreholeTWHC specimens for (a) wc ¼ 6%, (b) wc ¼ 7%, and (c) wc ¼ 8%, and the effect of theconfining pressure on the strength of the specimens.

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531528

As shown in Fig. 13a, unlike the solid specimen test results, theTWHC specimens did not exhibit strain hardening behaviour afterthe maximum strength and the peak strength can be determined inthe applied stress ranges. Fig. 13a shows that the ductility of thespecimens increases with an increase in the confining pressure as isthe case for the solid specimens.

Also, increasing the cement content by a small amount slightlyincreases the stiffness of the specimens for the same samplepreparation and test condition.

The results show that the confining pressure has a minimal ef-fect on the stiffness as exhibited in the stressestrain curves. Thisresult agrees with that of Mogi (2007) who showed that theyielding stress of ductile rocks does not increase with increasingconfining pressure.

5.4.1. Effect of wc and sconf on lateral and volumetric strainsFig. 11 also shows the deviatoric stress versus the average lateral

strain for different values of wc and confining pressures. Analysis ofthe micro-camera video recordings and Fig. 11 show that, for agiven wc, with an increase in the confining pressure, the boreholebreakout initiates at a lower lateral strain and the ductility in thedirection of sconf decreases. However, as mentioned in the previoussection, increasing the confining pressure results in greaterductility.

Fig. 10 also shows that, at lower confining pressures (i.e. 1 MPaand 2 MPa), the TWHC specimens transit to the dilation mode andwith an increase in the confining pressure the borehole breakoutinitiates in a contraction mode as in solid specimens. This is mainlydue to the fact that high confining pressures create boreholeconvergence and therefore the size of the sample in the lateraldirection reduces. Also, due to the presence of a borehole at lowconfining pressure, the volumetric strain curve conjugates thehorizontal axis at higher axial strain value in comparisonwith solidspecimens in the same confining pressure, suggesting that micro-cracks are developing on the borehole wall and sand grain dislo-cation occurs in the TWHC specimens versus solid specimens torelease the applied stresses and return the equilibrium condition tothe specimen. In addition, as shown in Fig. 10aec, with an increasein wc, the lateral strain increases compared to the axial strain andprior to borehole breakout in the TWHC specimen, axial contrac-tion of specimens with higher wc is less than that of the specimenswith lower cement content. In other words, an increase in wc re-sults in decrease in the pore spaces of the specimens, and appli-cation of the axial deviatoric stress causes less contraction in theaxial direction. Therefore, for higher wc values, the lateral strain ofpoorly cemented sand specimens is more dominant in comparisonwith the axial contraction. Based on the real-time camera re-cordings, it can be stated that, in the samples where dilation failurehappened, the transition from the volumetric strain contraction tothe dilation occurred before the borehole breakout initiation. Alsofor the same specimen condition, increasing the cement value by asmall amount slightly increases the stiffness of the specimens un-der a certain confining pressure (Fig. 13b).

5.4.2. Size-scale effect on TWHC specimensAs mentioned above, the TWHC specimens were prepared in

two borehole sizes: 10 mm and 20 mm. The same process wasadopted for the 20 mm borehole specimens, including grain sizedistribution, curing time, compaction force, and so on. Triaxial testswere performed on 20 mm borehole specimens for different wcvalues and confining pressures. Fig. 14a presents the deviatoricstress versus the axial and lateral strains for 20mmborehole TWHCspecimens for wc ¼ 7% and different confining pressures. Fig. 14billustrates the results of the specimens with 10 mm and 20 mmborehole sizes to compare their behaviours. As expected, thestrength of the 20 mm borehole TWHC specimens at failure isgenerally lower than that of the smaller borehole specimensespecially at higher confining pressures due to size-scale effectwhich was shown by Carpinteri (2002). Also, it shows a consider-able decrease in ductility for the 20 mm borehole specimens versusthat for the specimen with a 10 mm borehole. Fig. 14b shows that

Fig. 12. Sand dislocation and borehole collapse processes due to the hydrostatic pressure with high confining pressure (6 MPa) prior to the application of the deviatoric stress to theTWHC specimen.

Fig. 13. (a) Diagram comparing stressestrain behaviour of 10 mm borehole TWHC and solid specimens for wc ¼ 6%; and (b) Effect of cement content on the strength of the TWHCspecimens for a given confining pressure (sconf ¼ 3 MPa).

Fig. 14. (a) Deviatoric stress versus axial and lateral strains of 20 mm borehole TWHC specimens for wc ¼ 7%; and (b) Diagram comparing stressestrain behaviours of 10 mm and20 mm borehole specimens for wc ¼ 7%.

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 529

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531530

the stiffness of the TWHC specimens does not change withincreasing borehole size. This is in agreement with Mogi (2007)who suggested that the stiffness depends on the rock materialand, since the same mixtures were used for both specimens withdifferent borehole sizes, the same stiffness was observed from thetests.

In lateral direction, Fig.14a shows that with increasing confiningpressure, the lateral strain decreases for specimens with 20 mmborehole and the same wc values. In addition, it was observed thatunder different confining pressures the lateral strain at the bore-hole breakout initiation point is considerably lower in the 20 mmborehole specimens when compared with the 10 mm boreholespecimens (see also Fig. 11b). However, the lateral stiffness remainsunchanged and is unaffected by the increase in borehole size from10 mm to 20 mm. Also, the lateral ductility significantly decreasedin the specimens of larger borehole size. In other words, thebreakout in the borehole with larger diameter (20 mm) occurred ata lower strain compared to that for a smaller diameter borehole(10 mm).

6. Conclusions

This study has examined the stability of boreholes in poorlycemented sand by a series of newly developed laboratory tests. Itwas observed that the strength of the poorly cemented sandspecimens is largely influenced by the mould conditions. The real-time monitoring of preliminary tests on TWHC specimens showedthat in the case of curing the specimens for less than 8 days(ww ¼ 10%), borehole breakout initiated in a zone which was notfully dry. At a given wc, the peak strength and stiffness of thespecimens increased with an increase in the weight of coarse sandgrains. Also, the specimens showed more brittle behaviour forhigher levels of d.

The post-peak stressestrain behaviour of the specimens withlower wc showed more ductile trend and the effect of wc was moreconsiderable than that of d in the UCS tests. In lower wc values,dilation occurred after a limited compression in the specimens andwith the increasing cementation the lateral strain decreased andthe volumetric strain was mostly dominated by the axialcompression strain.

Using the videos recorded during the preliminary test on TWHCspecimens, the strengthening effect of the applied coating wasinvestigated and it was found that the borehole breakout started inthe area out of the section where polymer was applied.

Solid specimens exhibited strain hardening behaviour and nopeak strength was observed in their stressestrain diagram both inaxial and lateral directions. Unlike the solid specimens, TWHCspecimens did not exhibit strain hardening behaviour after themaximum strength was reached and the peak strength could bedetermined in the applied stress ranges.

At lower confining pressures (i.e. 1 MPa and 2 MPa), solidspecimens started contracting initially from the top and bottomsections and after 0.5%e0.6% of axial compression strain, it divertedto the lateral dilation. Increasing the confining pressure kept thespecimen in contraction mode and dilation started with delay at ahigher axial strain. For confining pressures of more than 2 MPa,specimens remained in contraction mode and never diverted to thedilatation until the maximum strength was reached.

For the TWHC specimens with confining pressures higher than4.5 MPa, the first stage of the test could not be completed and theborehole failed before the application of the deviatoric stress. Also,in the TWHC specimens the volumetric strain curve conjugated theaxial strain axis at a higher magnitude which suggested that micro-cracks were formed on the borehole wall and sand grain dislocationtook place in TWHC versus solid specimens to release the applied

stresses. Based on the observations from the real-time camerarecording, it can be stated that the transition from the volumetricstrain contraction to the dilation occurred before the initiation ofborehole breakout.

The failure strength of the 20 mm borehole TWHC specimenswas less than that of the 10 mm ones especially at higher confiningpressures due to the size-scale effect. Also, ductility was less in20 mm borehole specimens in comparison with that in the 10 mmones. However, the stiffness in axial and lateral directions did notchange with an increase in the borehole size. Also, it was observedthat the lateral ductility significantly decreased in the specimenswith larger borehole size.

For further investigation on the behaviour of poorly cementedsands, it is suggested to upgrade the laboratory tests to polyaxialstress condition in order to simulate anisotropic horizontal in-situstresses adjacent to a drilled borehole.

Conflict of interest

The authors wish to confirm that there are no known conflicts ofinterest associated with this publication and there has been nosignificant financial support for this work that could have influ-enced its outcome.

Acknowledgements

This work has been supported by the Deep Exploration Tech-nologies Cooperative Research Centre whose activities are fundedby the Australian Government’s Research Programme. This is DETCRC Document 2015/262.

References

Adeyanju OA, Olafuyi OA. Experimental studies of sand production from uncon-solidated sandstone petroleum reservoirs in Niger-Delta. Nigerian Journal ofTechnology 2011;30(2):18e30.

Al-Ajmi AM, Zimmerman RW. Stability analysis of vertical boreholes using theMogieCoulomb failure criterion. International Journal of Rock Mechanics andMining Sciences 2006;43(8):1200e11.

Al-Awad MNJ, El-Sayed AH, Desouky SEM. Factors affecting sand production fromunconsolidated sandstone Saudi oil and gas reservoir. Journal of King SaudUniversity, Engineering Sciences 1999;11(1):151e74.

Alsayed MI. Rock behaviour under multiaxial compression. PhD Thesis. Newcastle,UK: University of Newcastle upon Tyne; 1996.

Ashby MF, Jones DRH. Engineering materials 2: an introduction to microstructures,processing and design. Elsevier; 2014.

Bridgman PW. Studies in large plastic flow and fracture with special emphasis onthe effects of hydrostatic pressure. New York-London: McGraw-Hill; 1952.

Carpinteri A. Size-scale effects in the failure mechanisms of materials and struc-tures. CRC Press; 2002.

Chang C, Moos D, Zoback MD. Anelasticity and dispersion in dry unconsolidatedsands. International Journal of Rock Mechanics and Mining Sciences1997;34(3e4):48.e1e48.e12.

Chen X, Wu S. Influence of water-to-cement ratio and curing period on porestructure of cement mortar. Construction and Building Materials 2013;38:804e12.

Cleary MP, Melvan JJ, Kohlhaas CA. The effect of confining stress and fluid propertieson arch stability in unconsolidated sands. In: SPE Annual Technical Conferenceand Exhibition. Las Vegas; 1979. Nevada, SPE 8426.

Ewy R, Cook NGW. Deformation and fracture around cylindrical openings in rockdI.Observations and analysis of deformations. International Journal of Rock Me-chanics and Mining Sciences & Geomechanics Abstracts 1990a;27(5):387e407.

Ewy R, Cook NGW. Deformation and fracture around cylindrical openings inrockdII. Initiation, growth and interaction of fractures. International Journal ofRock Mechanics and Mining Sciences & Geomechanics Abstracts 1990b;27(5):407e27.

Ewy RT, Cook NGW, Myer LR. Hollow cylinder tests for studying fracture aroundunderground openings. In: Proceedings of the 29th US Symposium on RockMechanics (USRMS). Minneapolis, MN: American Rock Mechanics Association;1988.

Fischer PW, Pye, DS, Gallus JP. Method for drilling a well through unconsolidateddolomite formations. Patent US4120369; 1978.

Geertsma J. Some rock-mechanical aspects of oil and gas well completions. Societyof Petroleum Engineers Journal 1985;25(6):848e56.

S.S. Hashemi et al. / Journal of Rock Mechanics and Geotechnical Engineering 7 (2015) 519e531 531

Gough D, Bell J. Stress orientations from borehole wall fractures with examplesfrom Colorado, east Texas, and northern Canada. Canadian Journal of EarthSciences 1982;19(7):1358e70.

Gueguen Y, Palciauskas V. Introduction à la physique des roches. Hermann. 1992(in French).

Hagin PN, Zoback MD. Viscous deformation of unconsolidated reservoir sands. Part1: time-dependent deformation, frequency dispersion, and attenuation.Geophysics 2004;69(3):731e41.

Haimson BC, Song I. Laboratory study of borehole breakouts in Cordova Cream: acase of shear failure mechanism. International Journal of Rock Mechanics andMining Sciences & Geomechanics Abstracts 1993;30(7):1047e56.

Hall JCD, Harrisberger WH. Stability of sand arches: a key to sand control. Journal ofPetroleum Technology 1970;22(7):821e9.

Hashemi SS, Momeni AA, Melkoumian N. Investigation of borehole stability inpoorly cemented granular formations by discrete element method. Journal ofPetroleum Science and Engineering 2014a;113:23e35.

Hashemi SS, Taheri A, Melkoumian N. Shear failure analysis of a shallow depthunsupported borehole drilled through poorly cemented granular rock. Engi-neering Geology 2014b;183:39e52.

Hoskins ER. The failure of thick-walled hollow cylinders of isotropic rock. Inter-national Journal of Rock Mechanics and Mining Sciences & Geomechanics Ab-stracts 1969;6(1):99e125.

Jaeger JC, Cook NGW, Zimmerman RW. Fundamentals of rock mechanics. JohnWiley & Sons Inc.; 2007.

King LV. On the limiting strength of rocks under conditions of stress existing in theearth’s interior. The Journal of Geology 1912;20(2):119e38.

Kongsukprasert L. Time effects on the strength and deformation characteristics ofcement-mixed gravel. PhD Thesis. Tokyo, Japan: University of Tokyo; 2003.

Kongsukprasert L, Tatsuoka F, Tateyama M. Several factors affecting the strengthand deformation characteristics of cement-mixed gravel. Soils and Foundations2005;45(3):107e24.

Lee M, Haimson B. Laboratory study of borehole breakouts in Lac du Bonnet granite:a case of extensile failure mechanism. International Journal of Rock Mechanicsand Mining Sciences & Geomechanics Abstracts 1993;30(7):1039e45.

Mazanti B, Sowers G. Laboratory testing of rock strength. In: STP402: testingtechniques for rock mechanics. ASTM; 1966. p. 207e27.

Mogi K. Experimental rock mechanics. CRC Press; 2007.Obert L, Duvall WI. Rock mechanics and the design of structures in rock. New York:

John Wiley & Sons Inc.; 1967.

Paterson MS. Effect of pressure on stress-strain properties of materials. GeophysicalJournal International 1967;14(1e4):13e7.

Perkins TK, Weingarten JS. Stability and failure of spherical cavities in unconsoli-dated sand and weakly consolidated rock. SPE18244. 1988.

Pomeroy CD, Hobbs DW. The fracture of coal specimens subjected to complexstresses. Steel and Coal 1962;185:1124e33.

Robertson EC. Experimental study of the strength of rocks. Geological Society ofAmerica Bulletin 1955;66(10):1275e314.

Saidi F, Bernabé Y, Reuschlé T. The mechanical behaviour of synthetic, poorlyconsolidated granular rock under uniaxial compression. Tectonophysics2003;370(1e4):105e20.

Saidi F, Bernabé Y, Reuschlé T. Uniaxial compression of synthetic, poorly consoli-dated granular rock with a bimodal grain-size distribution. Rock Mechanics andRock Engineering 2005;38(2):129e44.

Tippieand DB, Kohlhaas CA. Effect of flow rate on stability of unconsolidated pro-ducing sands. In: Proceedings of the 48th Annual Fall Meeting of the Society ofPetroleum Engineers of AIME. Las Vegas; 1973. NV, SPE4533.

Dr. Saeed Hashemi is a Lecturer in the field of Geo-mechanics at CQUniversity in Australia. He received hisPh.D. from the University of Adelaide and his thesis was on“Drilling and maintaining stable boreholes in poorlycemented sands”. Saeed holds a M.S. degree in Geotech-nical Engineering from the Sharif University (the highestranked university in Iran), and a BSc degree in Civil Engi-neering from Iran University of Science and Technology.He is the author of several journal papers in Aþ and Ajournals (Excellence of Research Australia ranking). Saeedhas demonstrated experience in lecturing various geo-mechanics and rock mechanics courses and in supervisingstudent projects at the University of Adelaide and SharifUniversity. Besides his academic work experience, Saeed

also has more than 7 years of industry work experience where he worked on differentconstruction and oil and gas projects. Saeed received several research grants andawards during his career in Australia, such as Joint Research Engagement EngineeringCadetship Grant, Deep Exploration Technologies CRC, etc. Also, he is a permanentmember of Engineers Australia and Organisation for Engineering Order of Buildings inIran.